CoolFace
Datasetpublic

LocalLLaMA/local-model-explorer-data

Local Model Explorer Data An anonymous record of what people try to run locally, gathered by Local Model Explorer: the hardware they plan for (GPU memory, number of cards, system or unified memory), the models and context lengths they look at, which GGUF quants they open and copy commands for, and the llama-bench results and reports they choose to share. The question it answers: what hardware do local LLM users have, what do they try to run on it, and how fast does it actually… See the full description on the dataset page: https://huggingface.co/datasets/LocalLLaMA/local-model-explorer-data.

sourceHugging Facecc-by-4.0updated 45m agoView on Hugging Face
2likes1.3kdownloads
Dataset Card

Local Model Explorer Data

An anonymous record of what people try to run locally, gathered by Local Model Explorer: the hardware they plan for (GPU memory, number of cards, system or unified memory), the models and context lengths they look at, which GGUF quants they open and copy commands for, and the llama-bench results and reports they choose to share.

The question it answers: what hardware do local LLM users have, what do they try to run on it, and how fast does it actually go?

Quick start

python
from datasets import load_dataset

df = load_dataset("LocalLLaMA/local-model-explorer-data", "events", split="train").to_pandas()
df = df[df.suspicious_flags.apply(len) == 0]                      # clean rows only

# One row per visit: keep the newest summary of each session
visits = (df[df.event_type == "session"].sort_values("timestamp")
            .drop_duplicates("session_id", keep="last"))
visits.vram_class.value_counts()

# Community llama-bench results: generation speed per model, quant and GPU
bench = df[df.event_type == "llama_bench"]
bench.groupby(["selected_model", "quant", "gpu_name"]).generation_tps.median()

What is recorded

The dataset records outcomes, not clicks.

`event_type`One row perContents
sessionvisitHardware class, the filters the visit ended on, the top result and how it fit, and which models were opened, which repos were followed to the Hub and which run commands were copied. Re-sent as the visit changes: keep the newest row per `session_id`.
feedbackreport"Have you run this model?" with the quant, runtime, placement and how it went
llama_benchbenchmarkNumbers from a pasted llama-bench run, parsed by the Space

Schema (schema_version 1)

Every row has every column; columns that don't apply to an event type are null.

ColumnTypeRowsDescription
timestampstringallUTC, second precision
schema_version, app_versionint, stringall
event_type, session_idstringallsession_id is random per browser tab and discarded when it closes
hardware_kindstringallgpu, unified, cpu
gpu_modelstringallA listed card (e.g. NVIDIA RTX 3090) or custom
gpu_countintallCards
vram_classintallTotal GPU memory rounded down to a class: 0, 4, 6, 8, 10, 12, 16, 20, 24, 32, 40, 48, 64, 80, 96, 128, 160, 192, 256, 320, 384, 512 GB
ram_classintallSystem (or unified) memory rounded down to a class
unified_systemstringallApple Silicon Mac, AMD Ryzen AI Max+ 395 (Strix Halo), NVIDIA DGX Spark
hardware_sourcestringallpreset or custom
webgpu_vendorstringsessionWebGPU adapter vendor, e.g. apple, nvidia
browser_family, os_familystringsessionCoarse families
model_family, size_bucketstringsession, …Final filters; on reports, the model's
context, kv_typeint, stringsessionPlanned context length and KV cache precision (f16, q8_0, q4_0)
priority, sort, fits_onlystring, string, boolsession
result_count, families_searched, distinct_queries, searchedint, list, int, boolsessionWhat the visit looked at
top_model, top_quant, top_fitstringsessionFirst result and its fit: gpu, gpu_moe (experts in RAM), partial, cpu, unified, unified_tight, no
engine_versionstringsessionRanking engine, e.g. fit-v1
models_viewed, repos_clicked, commands_copiedlistsessionOpened models (max 10), repos followed (max 5), llama.cpp / ollama
selected_model, repo, quantstringreportsThe model (base repo id, or a name for variants), the GGUF repo and quant
quant_bucket, params_b, moe, arch, repo_downloads_at_selectionreportsFilled in by the Space from the catalogue
tried, outcome, quality_rating, failure_reasonstringfeedbackoutcome is worked or problem
runtime, placementstringfeedbackllama.cpp, ollama, lm-studio, …; gpu, experts_in_ram, partial, cpu, unified
reported_tps, reported_context, notesfloat, int, stringfeedbackNotes ≤ 280 characters; emails, links and phone numbers are removed
prompt_tokens, prompt_tps, generation_tokens, generation_tpsint, floatllama_benchFrom the pp and tg rows
backend, n_gpu_layers, n_threads, flash_attn, type_kllama_benchSettings reported by llama-bench
bench_model_type, bench_size_gib, bench_params_bstring, float, floatllama_benchWhat llama-bench says it loaded
gpu_name, buildstringllama_benchGeneric GPU name parsed from the output (e.g. NVIDIA GeForce RTX 4090) and llama.cpp build commit
suspicious_flagslistallEmpty when the row passed every check

suspicious_flags

FlagMeaning
unknown_model, unknown_repo, unknown_model_in_sessionNot in the catalogue at the time
repo_not_in_modelThe repo holds a different model
implausible_tpsFaster than memory bandwidth allows for that model (with 3× headroom)
benchmark_model_mismatchllama-bench's parameter count doesn't match the chosen model
unparseable_benchmark, incomplete_benchmark, incomplete_feedbackMissing numbers or the model
conflicting_feedback, implausible_hardwareContradictory or impossible values

Flagged rows are kept so filters stay transparent; the Space leaves them out of its statistics.

Layout and updates

data/incoming/YYYY/MM/DD/<instance>-<UTC time>-<suffix>.parquet   # written by the Space, at most hourly
data/events/YYYY-MM.parquet                                       # one file per finished month

Early each month the previous month is compacted into data/events/YYYY-MM.parquet (re-sent session rows reduced to the newest per visit) and its incoming files are deleted in the same commit. The events config reads both locations.

Privacy

Never collected: accounts, names, emails, IP addresses, cookies, user agents, device fingerprints, or individual clicks. Pasted llama-bench output is parsed on the server and discarded; only the numbers and settings above are stored, never file paths, CPU model strings or build machines. Hardware is stored as classes. Browsers sending Global Privacy Control or Do Not Track send nothing unless the user opts in.

Limitations

  • —Self-selected. Visitors are not a random sample of local LLM users.
  • —Planned, not owned. Hardware on session rows is what a visitor planned for, which may not be what they have. Benchmarks and reports are more reliable.
  • —Self-reported. Benchmark output can be edited before pasting; implausible values are flagged, not proven.

License

CC BY 4.0.